To evaluate the influence of the dose of co-trimoxazole prophylaxis on the risk of toxoplasmosis in human immunodeficiency virus (HIV)-infected patients, we performed a nested case-control study of 32 patients with toxoplasmosis (case patients) and 64 patients without toxoplasmosis (control patients) who were matched by CD4 cell count and Toxoplasma gondii serostatus; these patients were from a cohort of 521 HIV-infected patients who underwent a diagnostic neuroimaging study between March 1993 and January 1997. Twenty-seven (84.4%) of 32 case patients and 33 (51.6%) of 64 control patients received low doses of co-trimoxazole, a finding associated with an adjusted odds ratio (OR) of 9.36 (95% confidence interval [CI], 2.05-42.75) and indicating 89% protective efficacy for high doses. Fifteen (46.9%) of 32 case patients and 16 (25%) of 64 control patients were exposed to rifampin (adjusted OR, 3.38; 95% CI, 1.08-10.61). These results indicate that high doses of co-trimoxazole appear to be more effective than low doses for lowering the risk of toxoplasmosis in HIV-infected patients and that rifampin therapy may reduce the efficacy of co-trimoxazole.